Skip to main content
Dryad

Linking NDVI and leaf senescence in high arctic tundra plants: Soil moisture control on senescence phenology revealed by direct observations and NDVI sensors

Jørgensen, Andreas1; Semenchuk, Philipp2; Karlsen, Stein Rune3; Nilsen, Lennart1; Stien, Audun1; Cooper, Elisabeth1

Published Sep 11, 2026 on Dryad. https://doi.org/10.5061/dryad.6m905qgf1

Data files

Sep 11, 2026 version files 1.58 MB

Click names to download individual files

Abstract

Senescence phenology in Tundra plants is an understudied topic in plant ecology, limiting our understanding of trophic processes and vegetation-climate feedbacks linked to the autumn season in Arctic regions. In-situ Normalized Difference Vegetation Index (NDVI) sensors deployed at plot-level hold the potential to increase availability of quality data on senescence phenology by avoiding problems common to satellite-based remote sensing and intensive field campaigns in remote sites. We evaluated the relationship between plot-level NDVI and the progression of species level leaf senescence, as well as the suitability of NDVI midpoints as a metric to monitor community-wide senescence timing in tundra communities in high Arctic Svalbard. Ground validation data consisted of weekly observation of senesced leaf area (%) of 275 tagged plant individuals belonging to 11 species in six plant communities throughout the autumn 2023. Furthermore, we analyzed the influence of microclimate (soil moisture, soil temperature, and snowmelt date) on the timing cross-species senescence thresholds (25, 50, and 75 % senesced leaf area). Finally, we tested whether this analysis could be replicated using NDVI midpoints instead of field observations. We found that senescence progression was associated with a linear decrease in NDVI but the strength of the association varied between plant communities. NDVI midpoints occurred when community-weighted senescence had reached about 50 % (49 %, CI: [33 % - 69 %]) on average across all communities, making the NDVI midpoint a suitable metric for deriving meaningful end-of-season dates. NDVI midpoint occurred within 0-4 days of date when community-weighted senescence reached 50 %, excepts in graminoid-rich wetlands where the NDVI midpoint occurred about 12 days later. Out of the tested microclimatic parameters, soil moisture was the most important one. Here, all phenostages occurred later in wetter sites compared to drier sites. Furthermore, 25 % senescence occurred earlier in warmer sites. Date of NDVI midpoint similarly occurred later in wetter sites compared to dry sites. Using NDVI midpoints thus replicated the results based on field-derived phenostages. We conclude that plot-level NDVI midpoints are a promising metric for monitoring senescence phenology in tundra plants, with the potential to be used in future studies for investigating environmental drivers of senescence.